Microstructural parameters such as Triple Phase Boundary (TPB) density, specific surface area, connectivity and tortuosity of different phases strongly affect the performance of solid oxide fuel cells (SOFCs). Degradation of SOFCs is partially attributed to the coarsening of electrode during operation. The Potts Kinetic Monte Carlo (KMC) model [1], proven as a robust tool to study all stages of the sintering process, is an ideal tool to analyze the microstructural evolution of electrodes due to sintering [2]. The KMC model considers following three sintering mechanisms: (i) curvature driven grain growth by short-range atomic diffusion across the grain boundaries; (ii) pore migration by long range surface diffusion of pores; and (iii) vacancies formation at grain boundaries and annihilation along grain boundaries to cause densification. Hence, in the KMC models, the microstructures are controlled by sintering temperatures (K_BT) and frequencies (f), i.e., K_BT_gg, f_gg for grain growth, K_BT_pm, f_pm for pore migration, and K_BT_vf, f_vf for vacancy formation and annihilation.

As these sintering parameters are not material physical properties, we firstly calibrated these parameters in the KMC model by comparing the microstructural parameters measured from the simulated 3-dimensional (3D) microstructures and the FIB-SEM reconstructed 3D microstructures of La_0.6Sr_0.4Co_0.2Fe_0.8O₃ (LSCF) cathode pellet samples with different densities. The sintering parameters were calibrated using an artificial neural network (ANN) consisting of input layer (sintering material parameters), hidden layers (information processing units) and output layer (microstructural parameters). Inputs/outputs functional relationships were learned from a limited number of KMC simulations based on ‘Design of Experiment (DoE)’ based parametric studies. The trained ANN predicted more input/output pairs to form a database containing ‘material parameter-microstructural parameters’ pairs. Input sintering parameters that result in microstructural parameters within a desired error intolerance of the experimental results are called calibrated parameters. In this study, we identified input parameters as K_BT_gg=1.16, f_gg =0.68, K_BT_pm =1.48, f_pm =0.20, K_BT_vf = 37.0, and f_vf=0.52 for LSCF material. These calibrated parameters were validated successfully by predicting an extra experimental data, i.e. a further densified LSCF pellet sample.

By using the validated KMC simulations, we comprehensively examined the effect of the powder morphology on the microstructural characteristics of a La_0.6Sr_0.4Co_0.2Fe_0.8O₃ (LSCF) cathode. A number of numerical samples consisting of packing of spheres or clumped spheres were created using a discrete element method (DEM), taking into account the morphology of powders such as particle size, particle size distribution (PSD), aspect ratio (AR), sphericity, and particle orientation. The packings were subject to quasi-static compression under tri-axial pressures until having a prescribed relative density. Then, these DEM-generated numerical structures with different particle morphology were submitted to the KMC simulations. Their effects on relative density, specific surface area, and connectivity and tortuosity factors of LSCF and pore phases were compared and analyzed.

It is found in this study that using fine sized powder can improve the specific surface area and decrease the tortuosity factors of the LSCF phase. However, the decreased particle size at the same time enhances the sinterability of the LSCF, resulting in a decreased thermal stability of the LSCF cathode. The decreased particle size can also provide a decreased pore phase connectivity, which is disadvantageous for gas transport in the cathode. Secondly, a wider-PSD powder can improve the thermal stability by decreasing the sinterability of the powder. Another consequence of increasing width of PSD is that the pore phase tortuosity factor decreases, which benefits the gas transport in the cathode. The effect of the width of the PSD on the specific surface area and tortuosity factor of the LSCF phase is ignorable. Thirdly, powder with higher-AR particles has higher specific surface area and lower tortuosity factor for LSCF (better ionic and electronic conductivity). However, it also has a lower thermal stability and higher pore tortuosity factor. AR in the range of 1.0-2.0 seems has a very limited effect. Fourthly, powder with irregular shapes has better sinterability (poorer thermal stability). The tortuosity factors for LSCF and pores seems not significantly influenced in the study. Finally, in the case of elongated particles being used, alignment of particles in the green state will lead to lower sinterability of the packing, hence a better thermal stability of cathode. As a consequence of the particle alignment, anisotropic tortuosity factors for LSCF and pores have developed. In the alignment direction, the tortuosity factors appear smaller than the other orthogonal directions.

References

[1] G.N. Hassold, I-W. Chen, and D.J. Srolovitz, Journal of the American Ceramic Society, 73, pp. 2857-2864 (1990)

[2] S. Hara, A. Ohi, and N. Shikazono, Journal of Power Sources, 276, pp. 105-112 (2015)